Snippets Groups Projects

Something went wrong on our end

Forked from Scientific Computing / RStudio-Docu

11 commits behind the upstream repository.

5 years ago
155737fe
simplify ssh section · 155737fe
Dirk Sarpe authored 5 years ago
```
current windows versions support ssh via cli
```
155737fe

History
simplify ssh section
Dirk Sarpe authored 5 years ago
```
current windows versions support ssh via cli
```

Code owners

Assign users and groups as approvers for specific file changes. Learn more.

rstudio1.md 7.88 KiB

Short Notes documenting the RStudio server at iDiv

Subscribe to updates of this documentation via RSS.

Short Notes documenting the RStudio server at iDiv

Access Web-IDE of RStudio

check if you have been given access from GSU (i.e. if you are the group g_r_users)
go to https://rstudio.idiv.de
use your iDiv username and password to login
please remember to end your session when you have finished your current calculations
- interactively via the red button in the upper right
- in your longer running scripts use a statement like this after your computations. You might want to replace save.image() with save() if you are only interested in some result data and not the whole workspace.
```
# above some fancy computations
save.image('/data/myproject/myproject.RData')
quit(save = 'no')
```
if you want to run jobs for weeks or which require many cores (>10) for multiple days, please contact it-support@idiv.de beforehand. At this point you should consider to move to the HPC cluster.

ssh access

You can only reach the server from the internal network, i.e. if you are connected via VPN or cable at iDiv. Note that the internal name is rstudio1 instead of rstudio!

From a terminal/command prompt run (replace ab12cdef with your iDiv user name.

ssh ab12cdef@rstudio1.idiv.de

This even works on current windwos systems.

Data organisation

Currently there are three important places for you on the server:

/homes/$USER your home directory
/data place any larger files you want to work with here
/home/$USER/winhome a mounting point for your network home directory

Home directory

When you log in either via https or ssh this is usually the first place where you end up. Your home directory can be abbreviated by "~/". There is not a lot of space here, so do not place any larger data here. RStudio has the ugly not configurable habit to write quite some temporary data here (e.g. cached plots and suspended sessions), so please make sure that you clean up occasionally. It is a good place for your scripts.

RStudio stores several files like suspended sessions, graph history etc. in ~/.rstudio. This folder is purged every 90 days to remove "forgotten" suspended sessions and reduce wasted space. R itself stores workspace objects in .RData of your current working directory if you end your session or call save.image() without further arguments. Please ensure that you do not save large .RData files in your home directory.

/data

You can create your own directories here and place data inside. By default others can read the directories you created here, but not write in them. If you need any special permission let us know via it-support@idiv.de. To directly jump to another directory select the Files pane and click on the three dots (…) located on the right hand side of the current file path. Now enter the path you want to browse, e.g. /data/. Additionally there should be a symbolic link in your home directory called data, which points to the /data directory.

iDiv group shares

If you are part of an iDiv group you can access your group share on the rstudio server at /data/GROUPNAME-group-share/. If your group share is not available yet please contact it-support@idiv.de.

Network home

This is about \idiv.de\public\homes. To make it available on the rstudio server you need to get active. Log in via ssh or enter the pane labeled terminal in RStudio web IDE and run

su - $USER

You will be asked to enter your password again and afterwards your network home will be available at ~/winhome. However, it will only stay connected until you log out or quit your current session. If you have any longer running jobs you might want to consider either moving stuff from the ~/winhome to /data or open a tmux session from which you detach (Ctrl+b, d) before you disconnect.

R and data

compression

Please note that R is able to work with compressed files. This is especially useful if you are the typical csv/txt file user. Those files usually contain highly redundant data. Therefore compression can be very effective, e.g. the file which triggered me to write this was a txt file of 4 GB the gz compressed file had 98 MB. Many tools to read or write (e.g. save, save.image, read.table, fread from data.table) allow transparent use of compressed files, i.e. you just specify the compressed file instead of the uncompressed file.

Data transfer

You can use either use the web IDE to upload and export files.

Web IDE

In the Files pane click the Upload button to upload files to the current directory. For multiple files check the displayed TIP in the upload window. For big files please see below.

If you want to download/export files, select the checkbox for each file or directory and click More -> Export. If you selected multiple files a zip file will be downloaded.

Tools

Other file transfer tools are often more reliable and faster than the web IDE. Short instructions are available for several tools like scp, rsync, filezilla.

Code organisation

It is strongly recommended to use the version control system git to track changes in your code. It also helps you to distribute your code, be it for yourself (local machine, rstudio server, HPC cluster, …) or with others (e.g. via github or the iDiv gitlab). Ensure that you only track your code and not your data or results!

If you want your code to run in differenct environments (e.g. local machine, rstudio server or HPC cluster) be sure to separate environment specific code from your buisness logic. Recommendations and templates can be found in the EVE HPC Cluster wiki.

Packages

Many R packages are installed already via the system's package management and many more are available via the same track. This is the preferred way to install R packages, as it avoids duplicate installations per user, removes the need for manual package updates and helps that underlying libraries and packages are compatible with each other. Currently there are >3500 packages installed—check with installed.packages() for more details.

Of course you can still install packages via R's own functions (install.packages, devtools, …) if there is the need for it. Long term please drop a note to it-suppot@idiv.de so that we can install the package via the system.

https://github.com/eddelbuettel/rcppapt (not yet implemented)

Other servers with R

There are some other servers which have R and RStudio (desktop version) installed, but support for R on those is limited. This means you might frequently encounter outdated versions of R and associated packages.

idivgis01.idiv.de Biocon windows 2012r2 terminal server with gpu but restricted access
idivts6.idiv.de Biocon windows 2012r2 terminal server but restricted access
idivts7.idiv.de windows 2012r2 terminal server
idivts8.idiv.de windows 2012r2 terminal server

External resources available to you

Be aware that we do not control external resources and that you can connect your iDiv network shares directly.

Scientific Computing at the University of Leipzig

The UL department Scientific Computing makes a set of rstudio servers available as well. If you do not have an UL scientific computing account you can simply register one.